Case Studies

TitleDate RangeDescriptionHypotheses
Biscuit WildfireJuly 30 - August 1, 2002The Biscuit Fire was a wildfire that took place in 2002 that burned nearly 500,000 acres (2,000 km²) in the Siskiyou National Forest in southern Oregon and northern California. This event was chosen for Case Study analysis as an example of a relatively rare, large fire with high emissions and therefore a high potential of significantly affecting ozone formation.

The fire was highly active in the middle of ozone season (July-August). The analysis was initiated by looking at observed ozone within the vicinity of the fire, combined with CAMx modeling results, to look for exceedances of the 8-hour ozone NAAQS with a high estimated contribution from fire.

There were many NAAQS violations throughout California during this period, many of which had a contribution from fire. However, the analysis was limited to a geographic region in the vicinity of the Biscuit fire (even in this small region there are two ozone non-attainment areas and 11 Class I Areas). There were three ozone monitors with 8-hour ozone values >75ppb during the peak of emissions from the Biscuit fire, July 30-August 1, 2002. Two monitors exceeded the 8-hour ozone NAAQS on July 31, 2002, and had a large (>6ppb) contribution from fire predicted by the CAMx model. A summary of fire activity within the chosen study area shows a large amount of burning exclusively from wildfire; within the study domain were the Biscuit fire as well as two moderate-sized wildfires in southern Oregon. Examining the model animation, it was unclear whether the Biscuit fire contributed to observed ozone at both monitors. However, a HYSPLIT trajectory analysis it revealed that the regional meteorology was moving air due south, suggesting that the fire-caused elevated ozone observed in Jackson County, OR was due to the other wildfires burning at the time, whereas the Tehama County, CA monitor likely was influenced by emissions from the Biscuit fire (or all three fires).
McNally WildfireJuly 21 - August 29, 2002The McNally fire started in July of 2002 between Sequoia National Park and the Dome Lands Wilderness in southern California. The Biscuit Fire was also burning during this period, and smoke from the two fires merged over large swaths of California and Nevada.

Elevated ozone concentrations, including 8-hour averages above the ozone NAAQS, were measured consistently during this period throughout southern California. In many instances these exceedances would have occurred anyway, but the contribution from the McNally and Biscuit fires was significant throughout July and August. This case study highlights the complexity of assessing the impact from large wildfires amidst already elevated levels of ozone, including in several non-attainment areas. Understanding which exceedances qualify as Exceptional Events, and understanding how impacts from large wildfires affect SIP design, is a critical problem in areas with many sources of ozone-causing pollutants.

Attributing ozone impacts to a single fire event is difficult when there are several large events occurring simultaneously. Animations of CAMx modeling output and HYSPLIT trajectory analyses can help identify regions impacted by each event on a particular day. In this analysis, since all of the impacts from fire were due to wildfire emissions, it is less important to determine which fire caused the impact than to assess the total impact from fire on elevated ozone concentrations. The plots below showing observed 8-hour ozone combined with predicted fire contribution identify days where the fire impacts may have pushed ozone concentrations above the 8-hour NAAQS. There are also many instances where the observed ozone concentrations were high enough that an exceedance would have occurred regardless of the presence of fire.
Missionary Ridge & Hayman WildfiresJune 8 - July 2, 2002The Hayman and Missionary Ridge fires started in the summer of 2002 in Colorado, totaling over 200,000 acres burned. Concurrent emissions resulted in smoke transport and mixing across the region, and likely have contributed to ozone exceedances in Colorado and elsewhere.

Time-lapse animations of CAMx modeling results for 2002 show fire-caused ozone impacts across large swaths of the Southwest and Rocky Mountain regions originating from the Hayman and Missionary Ridge fires. Plots below show observed 8-hour ozone at several monitors with values above the NAAQS and a significant contribution from fire.
This case study demonstrates the potential of transported smoke from large wildfires to cause or contribute to exceedances of the 8-hour ozone NAAQS.
Northern California Wildfires, 2008June 20 - August 31, 2008The summer 2008 California wildfires, collectively dubbed the Northern California Lightning Series by CALFIRE, were a concentrated outbreak of wildfires during the summer of 2008. Over 2,780 individual fires were burning at the height of the period, burning large portions of forests and chaparral in California. The majority of the fires were started by lightning from dry thunderstorms on June 20.

An article published in Atmospheric Environment in 2013 examined median ozone concentrations at a monitor in Devil's Postpile NP in a "low-fire" year (2007) and a "high-fire" year (2008). A similar approach was taken below, examining observed maximum 8-hour concentrations paired with predicted contributions from fire from CAMx modeling at several monitors throughout California and Western Nevada. Boxplots of monthly observed ozone in 2007 and 2008 are shown for two monitors. The results corroborate the conclusions of the article that ozone levels were consistently higher in 2008 than 2007 during the period of intense burning in Northern California. Model results predicted, with varying accuracy, large contributions from fire to ozone concentrations on some days, as high as 30ppb. Model performance at each of the monitors highlighted show the model under-predicts more with decreasing latitude. The monitor in Tulare County experiences the heaviest influence of urban sources of ozone pre-cursors.
Rodeo-Chediski WildfireJune 21 - July 07, 2002The Rodeo-Chediski fire started in June of 2002, at the start of ozone season, and expanded rapidly throughout July. It burned 500,000 acres of Pondersa pine and shrubland, much of it with near-total consumption.
Exceedances of the ozone NAAQS were measured throughout the Southwest and into the Rockies during this period, and CAMx modeling results from our study show periods of significant ozone formation from Rodeo-Chediski and other fires burning simultaneously in the region.
Evans Road Wildfire (Pocosin NWR) / Peat burningJune 1 - August 1, 2008The Evans Road Wildfire was a smoldering peat fire in Eastern North Carolina that started on June 1, 2008. It burned 41,534 acres inside the Pocosin Lakes National Wildlife Refuge and smoldered for three months. It was chosen as a case study because of its persistence as a source and also to evaluate model and EI performance when dealing with peat fires.

CAMx modeling results, observed ozone data, and estimated fire emissions reveal a complicated scenario with implications for exceptional event designations, fire planning and SIP development. The map below shows over a dozen monitors that measured an exceedance of the 8-hour ozone NAAQS on 6/13/2008, during the peak of emissions as estimated by the fire emissions inventory. CAMx modeling results indicate a significant contribution from fire to ozone concentrations that day, but it is clear the model performed poorly during this period as it significantly overestimated ozone concentrations. The air chemistry in this part of the country during this time of year is complicated, with many anthropogenic sources of VOC and NOx combined with biogenic emissions. What is clear is that smoke from the Evans Road fire was present over the region in the days surrounding 6/13/2008, and may have influenced ozone concentrations significantly.

The overestimation of fire contribution to ozone in the CAMx modeling is partly due to the fire emissions inventory. Acres burned per day from the Evans Road fire were calculated using satellite-detected fire activity and scaled to the fire perimeter derived from MTBS data. If fire activity is detected in the same place over several days the emissions are partitioned over all days detected. In the case of Evans Road, the prolonged smoldering of the fire was likely not hot enough to detect via satellite, and therefore more emissions were compressed into a shorter time period. Peat fires have an extraordinary amount of available fuel to burn, as evidenced by the fire statistics summary below (compare acres burned to fuel consumed).

The fire contribution plots below show a small contribution from planned burns (agricultural and prescribed) that, while small, have the capability to influence ozone concentrations during the ozone season, which may be important for areas identified as marginal.
H1, H2
Mid-Atlantic Region 2008 Wildfire and Managed BurnsJune 25 - June 28, 2008The Evans Road Wildfire was a smoldering peat fire in Eastern North Carolina that smoldered for three months. Another case study looks at the Exceptional Event implications of this wildfire but also revealed a small impact from managed burns.
This case study uses the CAMx modeling results for 2008 to examine the relative contribution of different types of fire to ozone formation during a period in late June when impacts were the highest.
H1, H2
Fall burning in southern Louisiana, 2008September 27 - 30, 2008This case study was identified by interrogating the database of observed 8-hour ozone values in the "shoulder season" (Sep-Oct) in 2008 and looking for instances of elevated ozone. These results were then filtered to identify areas with planned fire-contributed ozone predicted by CAMx modeling.

This time period in southern Louisiana saw two days with ozone above the current 8-hour ozone standard with a significant fire influence. September 27th saw smoke transported from wildfires in Northern California and Canada carried southward, as can be seen from the HYSPLIT trajectory results and the modeled fire contribution plots (there was also a significant local influence from prescribed burning). Fire impacts on ozone were also seen on September 30th, but this time the impacts were predominately from nearby planned fires occurring throughout the state.
H1, H2, H3
Northern GA/southern TNOctober 3 - 6, 2008Open burning occurs in the Southeastern United States nearly year-round. This Case Study was identified by searching observed ozone data paired with CAMx modeling results from 2008 to find instances in the "shoulder seasons" with elevated ozone attributed to fire. October 4, 2008, saw an exceedance of the ozone NAAQS at a monitor in Chattanooga, TN, with a small, but significant, estimated contribution by fire. The fire summary map reveals that planned fires of different types were occurring throughout the region. CAMx modeling results predict that, despite the "obvious" source of agricultural burning from the west, the largest contribution was from nearby prescribed burning.H1, H2, H3
Agricultural fires' influence on ozone formation - MNApril 14 - 15, 2008April is a time of intense agricultural burning in the Midwest, especially in Kansas. Over a period of 2 days, April 14-15, 2008, an estimated 16,000 acres of agricultural land was burned. CAMx model results were interrogated to look for instances of fire contributing to ozone formation at ozone monitors in the Midwest during this period. Three monitors in Minnesota and Wisconsin were modeled to have a nearly 5ppb contribution from agricultural burning. The map below shows minimal burning in the vicinitiy of the western-most monitor in Minnesota, so presumably the influence is from the burning in Kansas to the south. Measured ozone concentrations, while not exceeding the current 8-hour ozone standard, were elevated enough such that a lowered standard could cause greater concern about smoke transport to this region in the Spring.H2, H3
Annual burning patterns in the Southeast United StatesOctober, 2004 - 2007This is one of three case studies that looks at patterns of burning over several years. Burning in October 2008 in the Southeast US was analyzed as a separate case study and showed the potential for fire to contribute to elevated ozone in the Fall. This analysis looks to see if similar patterns of burning occur in other years or if 2008 was unusual.

Daily fire activity estimated by the FETS from 2004 - 2008 is displayed, and reveals a consistent pattern of burning from October to April each year, with a minimum in the summer (the spike in May 2007 was due to the Big Turnaround Complex, the largest fire in Georgia history). The annual pattern of fire activity does not clearly correlate with annual trends in observed ozone, seen by comparing tons consumed to observed ozone at a monitor in Hamilton County, Tennessee. However, looking at the month of October from 2005-2007, tons consumed in the study area on and/or leading up to days of elevated ozone observed at the Hamilton County monitor.
H2, H3
Chatfield, CO July 2004-2007July, 2004 - 2007Chatfield, CO July 2008 was analyzed as a separate case study looking at fires' contribution to ozone formation in a Non-Attainment Area (NAA). That case study concluded that smoke from wildfires burning throughout the Western US contributed to elevated ozone just above the 8-hour ozone standard at a monitor in Douglas County, CO. This case study pairs observed ozone data in non-model years with fire activity in the same geographic domain. While it is difficult to conclude that fire contributed to ozone exceedances in those years, there are instances, similar to 2008, where a spike in fire activity lead to a period of elevated ozone above the 8-hour standard several days later.H2, H3
Chatfield, CO July 2008July 4 - 18, 2008Denver, CO is a non-attainment area for ozone and has a mixture of sources contributing to ozone formation. Ozone monitors during the summer months often hover just below the ozone NAAQS. Introducing fire emissions into the mix has the potential to bump concentrations just over the standard that otherwise would have been just below.
This case study was identified by filtering observed ozone data paired with CAMx modeling results to find NAAQS exceedances just above the standard with a significant contribution from fire. Burning was occurring during this time period to the east (Midwest Ag fires) and to the west (large wildfires in Northern California). The source apportionment from CAMx clearly shows the signal is dominated by wildfires, and therefore the emissions are coming from the Western US.
H2, H3
Fall burning patterns in southern Louisiana, 2004 - 2007Sep - Oct, 2004 - 2007This is one of three case studies that looks at patterns of burning over several years. Managed burning in early Fall 2008 in southern Louisiana was analyzed as a separate case study and showed the potential for fire to contribute to elevated ozone in the shoulder season. This analysis looks to see if similar patterns of burning occur in other years or if 2008 was unusual.

Pairs of box plots and fire activity tables for each year from 2004 - 2007 are displayed, and reveal a consistent pattern of managed burning (mostly prescribed fire) in the September-October time-frame. A time series of tons consumed for the entire 4 year span show an annual pattern of burning mainly in the spring and fall. The ramp-up in burning in Sep-Oct coincides with sporadic instances of elevated ozone, whereas in the summer ozone is high with little burning occurring. Two years, 2005 and 2007, have several larger fires burning simultaneously for short periods but that do not show increased frequency of elevated ozone. Two monitoring sites from the area show instances of elevated ozone into October; the site at LaFourche Parish has a more consistent annual pattern. The ozone threshold chosen for this analysis was 70ppb.
H2, H3
Flint HillsApril 2008Each year in April, thousands of acres of agricultural land are burned in the Flints Hills region of Kansas. April is generally considered a "shoulder" season for ozone formation, and this annual event was chosen as a Case Study to examine its effect on ozone formation throughout the region.
Observed ozone data in the vicinity was examined for the month of April, and compared with CAMx model results to show the estimated impact of fire on observed values. Fire emissions data for non-model years were summarized to demonstrate the persistent nature of burning in this region each year.
H2, H3
Managed burning proximate to Non-Attainment AreasApril - October, 2008The fire emissions inventory developed for 2008 revealed several ozone non-attainment areas (NAAs) that have planned fires inside or near the NAA boundary at certain times of the year. Burning proximate to NAAs has the potential to affect ozone formation in areas already prone to elevated ozone.

Three NAAs were isolated and observed 8-hour ozone concentrations were compared to CAMx modeling results that predict the contribution by 3 types of burning to ozone formation. Three NAAs were chosen by ranking the total tons consumed by agricultural and prescribed burning (together referred to as 'planned fires') within 100km of the NAA boundary between April and October, 2008. Timeseries plot of one ozone monitor representative of each area are presented below with modeled fire contributions. Reference maps for each area are also presented.

The modeled fire contribution plots show that the majority of impacts from fires at each ozone monitor is caused by wildfire in the summer months, and by planned burning in the spring and fall. The monitor in Arizona has a distinct annual trend such that spring and fall ozone concentrations do not approach 70ppb, but both Polk County, Texas and Dawson County, Georgia see periods of elevated ozone in the Spring and Fall coincident with planned burns.
H2, H3
Prescribed fires' influence on ozone formation - AZSep 30 - Oct 2, 2008Prescribed burning occurs in Arizona in the Fall. As seen in the map below, some of this burning occurs near the nonattainment area (NAA) in the Phoenix Metro Area. October is generally not a problem for ozone exceedances in this area, but the case study below shows that prescribed burning has the potential to contribute to ozone formation during this time, and a lower standard of 65ppb may require more careful planning in the shoulder seasons in this area, as revealed by the plot below of observed ozone and model fire contributions at a monitor in Maricopa County, inside the NAA.H2, H3
Prescribed fires' influence on ozone formation - MTOct 12 - 25, 2008This case study was developed by looking at areas with large amounts of prescribed burning and coincident prescribed fire-caused ozone predicted by CAMx modeling in 2008. Idaho and western Montana have an active prescribed fire season in the fall, as can be seen in the map below that covers a two-week period in October, 2008. One ozone monitor location still active during this time period in 2008 was shown to have a significant amount of predicted ozone formation caused by prescribed fire. However, observed ozone at this site was quite low, and closer examination of the model performance shows consistent over-prediction of total ozone, shown below. Despite the model over-prediction, estimated values with up to 5ppb of prescribed fire-caused ozone are still well below the current or any potential near-term new standard. So, while there is evidence that prescribed burning in this area at this time can influence ozone concentrations, it is not enough to be a concern in the context of fire planning or SIP development.H2, H3